Style and Configuration in Prehistoric Iconography

The iconography of ancient art has to do with making propositions about what that art depicts. Stylistic studies, in contrast, make propositions about the sharedness of formal properties with other objects. These can be done separately. Although many iconographic analyses of ancient art proceed with little or no consideration of style, I argue that the two modes of analysis are interdependent. I offer a methodological case that stylistic analysis is logically prior to iconography study in the domain of ancient art. This observation argues for a distinctively staged approach to the iconography of ancient objects, one in which detailed stylistic study is a necessary prerequisite to success in determining the referents of representations. In this chapter, I wish to make what is really a fairly simple methodological point concerning the relationship between the iconography of ancient art and the study of style. I think the point is worth making because I have found that there is far less consensus about method among prehistoric iconographers than many would readily admit. I argue that stylistic and iconographic analyses are interdependent and that stylistic analysis is an indispensable prerequisite to a more confident, successful iconography. This suggests, in turn, a staged methodology incorporating both. V E R N O N J A M E S K N I G H T J R . 224 Before going any further, let me situate myself, so to speak, in order to reveal some of my biases. In the New World, pre-Columbian iconography of art is divided between two sets of practitioners: one trained in the field of art history and a somewhat larger group trained as anthropologists. These two groups, however crisply divided by their training, nevertheless do interact extensively and borrow methodologically from one another. Personally, by training I am an anthropologist, one who works in eastern North America and the Greater Antilles. Further, I am one of the original members of the Mississippian Iconographic Workshop, originally a spinoff of Linda Schele’s Maya Hieroglyphic Workshop, which has met annually since 1993 in Texas. At the moment, the group is responsible for three edited volumes (Lankford, Reilly, and Garber 2011; Reilly and Garber 2007; Townsend and Sharp 2004) and a number of additional articles. My interest in methods is fresh in that I spent the greater part of 2010 working on a book manuscript on the subject (Knight 2013), having received a stipend for that purpose by Dumbarton Oaks in Washington, DC; I have taught the subject as a graduate seminar for twenty years. As for my theoretical perspective, it comes ultimately from a background in symbolic and structural anthropology acquired during the 1970s. My view of the nature and role of culture has a great deal in common with cognitive anthropology as outlined by D’Andrade (1987) and others. In short, I view both style and iconographic communication as governed by schematic cultural models of differentially distributed knowledge. Iconography is fundamentally about the relationship between representational images and whatever they refer to, outside of themselves. I use the term referent intentionally, as opposed to the more slippery term meaning. Iconography is therefore completely different from the study of style (Panofsky 1939), which has to do with S T Y L E A N D C O N F I G U R AT I O N I N PR E H I S T O R I C I C O N O G R A PH Y 225 cultural models governing the manner in which images are depicted. I further restrict myself to the iconography of prehistoric images. These are images to which any contemporaneously written record is completely denied. The subject matter therefore includes imagery produced by such pre-Columbian groups as the Olmec, Teotihuacán, and Izapa of ancient Mexico, Coclé of Panama, the Marajoara of the Amazon Basin, the Chavín and Moche of Peru, and many others. These ancient peoples, all of whom left a wonderful variety of art for our modern contemplation, have in common that they were organized as what most archaeologists would call “complex societies,” meaning simply that they were socially differentiated to some degree, and their communities were politically organized. I exclude such groups as the Classic Maya, who had a well-developed glottographic writing system. In Classic Maya art, imagery is often combined directly with text so that the burden of communication is shared. This creates entirely new genres (e.g., Berlo 1983), and in a very real sense, it changes the rules of iconographic interpretation. Further, iconography can be considered a subset of cognitive archaeology, although our practice diverges very much from standard descriptions of what cognitive archaeology is by such worthies as Colin Renfrew (1994), Kent Flannery, and Joyce Marcus (1998). From the art historical perspective, what we are doing could be considered a facet of what Erwin Panofsky (1939) called “iconology,” although Panofsky himself might not recognize it as that, were he alive today. I claim that where ancient images are at stake, there is a special methodological relationship between the study of styles and the study of representations. Let me introduce the matter with a paradox. In North and South America respectively, among the most profound sources of iconographic insights published to date are two research projects. First, for North America, is Philip Phillips’s and James Brown’s (1975-1982) magnificent six-volume set concerning V E R N O N J A M E S K N I G H T J R . 226 the corpus of Mississippian engraved shells from the Craig Mound at the Spiro site in Oklahoma. For South America, the corresponding work is the extraordinarily influential two-volume set on Moche fineline painted ceramics by Christopher Donnan, Donna McClelland, and Donald McClelland (Donnan and McClelland 1999; McClelland, and Donnan 2007). These two studies have much in common, including the sheer amount of labor that went into collecting the corpus in both, a methodological imperative inherited from art history (Kubler 1967, 1969). The Spiro shell volumes brought together 791 artifacts, presented in a common format, as rubbings gathered from numerous scattered collections by a team of four artists working over a six-year period. The Moche fineline pottery database is even more impressive in this regard. It is a photographic archive documenting over 2,300 pottery vessels from museums and private collections worldwide assembled over a period of thirty years. As many as twenty to thirty photographs of each vessel were taken. Because these paintings were most often done in the round, for the purpose of analysis and publication, the photographed paintings were converted into inked two-dimensional rollouts. The paradox concerning these volumes is as follows. Although both the North and South American studies have been extraordinarily fertile sources of iconographic interpretation in the years since their publication, neither study mainly concerns iconography. Instead, they are fundamentally stylistic studies, which subdivide their materials into style groups and style phases using formal traits of execution and layout. Their concern with the subject matter of the art is secondary in both cases, and its discussion develops only after sufficient control of style and stylistic change over time has been achieved. Both studies are extraordinarily conservative in regard to interpretation of the imagery, especially in their shared disavowal of any sort of ethnographic analogy to help interpret what is being depicted. S T Y L E A N D C O N F I G U R AT I O N I N PR E H I S T O R I C I C O N O G R A PH Y 227 The importance of this is that many of the attitudes expressed in these volumes are at odds with dominant tendencies among practicing prehistoric iconographers. Many ignore the imperative to collect a full corpus of a single genre before proceeding. Instead, they are content to move interpretively from object to object, skip from genre to genre, and follow alleged motifs or themes across great distances in time and space. They tend to employ ethnographic analogy promiscuously, using documentation from the ethnographic present that they project backwards in time with little regard for the possibility that forms might become disjoined from their subject matter over time. In doing so, these practitioners appear satisfied to interpret ancient images merely as illustrations of already known ethnographic concepts. More to the point of this chapter, they tend to shun the laborious work of stylistic analysis, preferring to jump headlong into the arena of iconographic subject matter. They tend to assume the existence of styles, especially so-called “international styles” such as the Olmec or Teotihuacano, where there has been no such formal demonstration. Even a casual look at such “styles” reveals enormous diversity and complexity. I believe these shortcuts are a mistake and that detailed stylistic analysis is methodologically essential as a precursor to practicing iconography with ancient images. Before zeroing in on this relationship more precisely, I need to say just a bit about methodology in iconographic work itself. In simple terms, like stylistic analysis, it begins with a collected corpus of works of the same time period and ideally of the same genre. Insistence on working within one genre at a time—say engraved shell cups or fineline painted pots—echoes what the art historian Ernst Gombrich (1972) called the “principle of the primacy of genres,” which says that representations change as genres change. To take an extreme example, a representation of the Greek goddess Athena on a coin is not to be compared with the image of Athena as the V E R N O N J A M E S K N I G H T J R . 228 central object of devotion in the Parthenon. The referents are very different at anything other than a superficial level. With a corpus of work assembled in a format favorable to comparison, analysis begins by deconstructing images into their parts, using a series of defined suprastylistic concepts, such as salient element, motif, filler motif, identifying attribute, classifying attribute, and ideograph. Most of these terms come directly from the vocabulary of art history, but with definitions refined and tailored over the years to the tasks at hand. By giving the discovered elements neutral names and by tracing their occurrences, relative positions, and clustering completely throughout the corpus, one arrives at sets of apparent common subject matter. Depending on the circumstance, these sets are labeled as visual themes and visual narratives and are again given neutral names so as not to bias the outcome of analysis. The art historian George Kubler (1967, 1969) called this method “configurational analysis.” Importantly, configurational analysis is conducted entirely without reference to ethnographic analogy (or more accurately, historical homology [Berlo 1983], since what is being compared are chronologically distant manifestations of the same cultural phenomena rather than merely analogous traits). Avoidance of historic documentation at this stage of analysis is deliberate. Kubler (1969) believed that historical disjunction between form and referent was such a persistent danger that complete avoidance was the prudent course. Nowadays, nearly all anthropologically trained iconographers, and probably most of the art historians as well, would not go that far. They would argue that historically recorded myths, rituals, beliefs, cosmologies, and so forth grant an indispensable foothold into the past. Methodologically then, the issue is to carefully control the application of these later sources, to minimize the possibility of simply reading the present into the past and irreversibly biasing the S T Y L E A N D C O N F I G U R AT I O N I N PR E H I S T O R I C I C O N O G R A PH Y 229 outcome. Past configurations are, to some degree, allowed to speak for themselves, being systematically compared with later information, noting both what seems to fit and what does not. Procedurally, the configurational analysis must both come first and must be analytically separate from any consideration of historical homologies. Because configurational analysis relies on internal information embedded within the images, it is fair to ask to what degree one can reliably say anything about subject matter and therefore do iconography. Art historians of Panofsky’s generation assumed that some kinds of subject matter recognition were universal, with factual understandings requiring no cultural knowledge. But a subsequent generation of theorists (e.g., Gombrich 1977; Hermerén 1969; Kippenberg 1987) pointed out the fallacy of factual recognition, even in art that we might think of as being naturalistic. So clearly, in images we see only what we are conditioned to see by shared cultural models—in this case, stylistic models of depiction. On this basis, Robert Layton (1977) has argued that there is no such thing as “naturalism” in art; all depiction is conventionalized. The one distinction that can be made is the degree to which the style in question is, or is not, obedient to perspective, in the sense of showing the contours of what the eye sees from a fixed vantage point. Some style systems ignore perspective almost entirely. Thus, our recognition of anything at all, iconographically, depends on how much of the stylistic model in question we happen to grasp. As in other aspects of iconography, success is largely a matter of knowing the context. My contention is that stylistic and iconographic analyses are separate but interdependent endeavors. I am arguing that an explicit grasp of the stylistic canons governing a corpus of related imagery is fundamentally prerequisite to any successful iconographic analysis of that corpus. To further drive home this point, let us consider three V E R N O N J A M E S K N I G H T J R . 230 things essential to prehistoric iconography that a study of stylistic conventions can tell us: what is what, what is contemporaneous with what, and what is local. Figure 9.1. Two winged serpents, from engraved pottery vessels from the Moundville site, Alabama ([a] from Moore [1907, figure 59]; [b] drawing by Erin E. Phillips). (Drawing and permission courtesy of Erin E. Phillips) First, an understanding of stylistic conventions can tell us what is what. Consider the two images given in figure 9.1, both drawn from the corpus of engraved pottery in the Hemphill style from the Moundville site in Alabama. Arguably, these Mississippian engravings depict the same theme, which is a winged serpent. The uppermost is a well-known illustration. It was published by Clarence B. Moore in 1907, and it has been occasionally reproduced in publications ever since. At times, this particular snake has been referred to in figure captions as a “plumed serpent,” with the objects on the head viewed as feathers. The unspoken comparison here is obviously with S T Y L E A N D C O N F I G U R AT I O N I N PR E H I S T O R I C I C O N O G R A PH Y 231 the “feathered serpents” of Mesoamerica. The lower image, in contrast, is commonly understood as “horned,” with the corresponding objects on the head seen as antlers. Again, there is an unspoken comparison, in this case with the horned underwater serpents of Native American myth in the eastern United States. The question is, does the upper configuration really depict plumes, or instead is it merely another convention for antlers? Fortunately, we have a comparative stylistic study of all thirtynine known winged serpent depictions from Moundville (Schatte 1997), and that study reveals the answer. It gives us empirical grounds to state with complete confidence that all such configurations on the head are to be interpreted as the same thing (antlers), which show a full range of schematization from fully to barely recognizable. It so happens that the lower example is chronologically early and the upper is late (E. Phillips 2011). Viewing images in isolation leads to incorrect conclusions, but systematic stylistic study of an entire corpus reveals a continuous range of conventional depiction. Figure 9.2 provides a second illustration of the idea, again taken from the engraved art on pottery from Moundville. I once gave a talk where I used this drawing to illustrate another of the common themes on Moundville pottery, which we call the crested bird. As I was explaining that this does, in fact, depict the head of a crested bird, a gentleman in the back of the audience raised an objection. While he saw that it could be a crested bird, he thought it much more likely to depict a fish in the process of consuming something large. He was seeing the hachures at the border between the beak and head as the teeth of a fish, and the crest as the fish’s dorsal fin. And frankly, if this were the only image we had of the subject, his interpretation could be as easily defended as mine. Obviously, we do not all necessarily see the same thing. In my case, if I had been armed at that moment with the full repertoire of engravings of the crested bird V E R N O N J A M E S K N I G H T J R . 232 theme, I could have explained the stylistic canons that govern its depiction. Because the remaining examples of the crested bird theme show nothing even remotely fish-like, my case could easily have been made. Even in cases where the analyst cannot identify the natural referent, stylistic analysis can still allow us to make informed choices about what is the same subject versus what is something else. Style informs the critical same-versus-different distinctions that make iconography possible. Figure 9.2. Detail from a depiction of a crested bird on engraved pottery from the Moundville site, Alabama (Moore 1905, figure 9). (Image in the public domain) Second, style helps us to decide what is contemporaneous with what. Stylistic seriation is an indispensable tool for arranging images in a chronology. The creation of style phases within traditions grants us “analytical moments” so to speak—synchronic units that give us a framework within which we can capture the local, potentially temporary, relations between images and their referents. Our recognition of style horizons also allows us to link contemporaneous works S T Y L E A N D C O N F I G U R AT I O N I N PR E H I S T O R I C I C O N O G R A PH Y 233 across larger geographical spans in order to consider whether their referents are the same or different. Conversely, a lack of attention to these details can lead only to jumbled comparisons of images drawn from different times and places, leading to false assumptions about continuity of forms and referents. In short, iconography is meaningless without the control of time, and style is a major contributor to the building of chronologies. Third, stylistic analysis can tell us what is local. Prehistoric complex societies were not isolated systems. In these societies, skillfully crafted goods were often sought from afar, or perhaps brought as gifts by emissaries seeking alliance. Nonlocal goods were ultimately distributed among both elites and non-elites, during feasts, marriages, funerals, and other consequential social events. As a result, the total assemblage of portable imagery at any given site, particularly at large civic-ceremonial centers, is a mix of goods produced locally and goods manufactured elsewhere. Among these nonlocal goods, typically there are images that not only are foreign to the local style but also bear subject matter that would have had no particular significance in the local context. In any effort to isolate local systems of images and referents, as a practical matter it is necessary to winnow out the nonlocal “noise,” simplifying the universe of images and eliminating from consideration much that would otherwise confound the analysis. Among the key tools for accomplishing this winnowing out of the foreign material is stylistic analysis (another being the chemical or geological sourcing of raw materials). Consider the two large stone effigy smoking pipes shown in photograph 9.1, depicting a supernatural panther. Although four such pipes have been found at the Moundville site in west Alabama, stylistically they are out of place. They do not belong to any of Moundville’s styles, but rather to a style known as Bellaire A that properly belongs in the Lower Mississippi Valley (Steponaitis et al. 2009), where many V E R N O N J A M E S K N I G H T J R . 234 more examples have been found. In confirmation of this, the limestone from which they are made has been sourced to outcrops near Vicksburg on the Mississippi River. Not only are the objects foreign at Moundville, but so is the subject matter. Of the many hundreds of locally produced images in several media at Moundville, none shows the long-tailed panther. It is a foreign subject but one that was probably reinterpreted according to local cultural models and incorporated into Moundville ritual practice in a limited way. Photograph 9.1. Limestone effigy pipes from the Moundville site, Alabama, depicting long-tailed panthers. (Photograph courtesy of Vincas Steponaitis) In summary, style, as pure form, can profitably be analyzed entirely apart from subject matter, and much work along these lines has been done. Nonetheless, the results of stylistic analysis have much to do with an understanding of referents. Style has the potential to inform us on what is what, what is contemporaneous with what, and what is local, all of which have a strong bearing on understanding suprastylistic configurations. In that sense, stylistic analysis is logically prior to configurational analysis and any considerations of S T Y L E A N D C O N F I G U R AT I O N I N PR E H I S T O R I C I C O N O G R A PH Y 235 iconographic reference. Aspects of iconography thus have a methodologically dependent relationship to aspects of style, just as configurational analysis of images is logically prior to the application of historic homologies. All of this argues for a distinctly staged approach to prehistoric iconography which is sometimes hinted at and other times carried out in practice, but is seldom explicitly laid out as I think it should be. I envision the ideal approach as having several stages: first, assembly of the corpus; second, organization of the material according to stylistic analysis; third, incorporation of natural history and archaeological field data; fourth, configurational analysis of suprastylistic units; fifth, careful application of ethnographic analogy; sixth, building of iconographic models; and seventh, testing these iconographic models. This last is because, after all, we do have to verify our claims.